LLMs can make user intent far more legible. The better you can prompt or describe your intent to a LLM, the better the output.